Existing　RDF　keyword　search　studies　focus　on　constructing　smallest　trees　or　subgraphs　which　contain　all　query　keywords,　but　neglect　the　semantic　association　between　RDF　data.　Thus,　this　paper　proposes　the　keyword　parallel　search　over　RDF　data　based　on　semantic　association　(KPSRSA))　algorithm　which　utilizes　a　score　function　to　measure　semantic　association　by　combining　OWL　ontology　and　the　probability　model.　It　uses　a　distributed　database　Hbase　as　a　storage　medium　and　Mapreduce　to　perform　parallel　query,　which　queries　sub　-clusters　with　semantic　association　in　Map　phase　and　constructs　a　series　of　associated　clusters　as　query　results　in　Reduce　phase.　The　experimental　results　demonstrate　that　the　KPSRSA　algorithm　improves　the　precision　and　relevance　of　search　results　and　keywords.　In　addition,　distributed　storage　and　parallel　computing　inquiry　has　improved　scalability.